Skip to content
Apply for the pilot
Planned

A server of your own, beside everything else.

Virtual machines with SSH access for the workloads you’d rather run yourself, and GPUs for inference once a GPU host is in place. Both are planned, and neither can be ordered until the hosting behind them is verified.

Apply for the pilotRead the docs
New serverPlanned
Size
1 vCPU · 1 GiB · 25 GB disk
1 vCPU · 2 GiB · 50 GB disk
2 vCPU · 4 GiB · 80 GB disk
4 vCPU · 8 GiB · 160 GB disk
8 vCPU · 16 GiB · 320 GB diskLater
Image
Ubuntu24.04 LTS
Ubuntu22.04 LTS
Debian13
Debian12
GPU instances
L4, one GPUWaitlist
L40S, one GPUWaitlist
Real virtual machinesYour own kernel and root shell, not a container dressed as a server.
SSH accessAdd your keys and administer the machine directly.
Ubuntu and DebianCurrent long-term support images to start from.
Sizes to fitFrom 1 vCPU and 1 GiB up to 4 vCPU and 8 GiB to begin with.
Lifecycle controlsStart, stop, reboot and rebuild from the dashboard or the API.
Persistent disksStorage that stays with the server through restarts.
Billed on their ownBy the hour, separate from your plan.
GPUs on a waitlistL4 and L40S instances for inference, once a GPU host is in place.

The rest of Dekka